Wireless Ad Hoc and Sensor Networks

Optimized Energy and Delay-Based Routing 359constraints imposed by the ad hoc networking environment. Therefore,any routing protocol proposed for a multihop ad hoc wireless networkmust consider the following design issues:Reactive vs. proactive routing approach: In the reactive routing approach(on-demand), a route is not identified unless it is required. Bycontrast, proactive protocols periodically exchange control messagesand provide the required routes instantly when needed. Inthe selection of reactive vs. proactive techniques, there is a tradeoffbetween the latency in finding the route to a destination andthe control traffic overhead.Centralized vs. distributed approaches: Because there is no centralizedcontrol in wireless ad hoc networks, a distributed routing protocolis preferred.Optimal route: The definition of an optimal route is very importantfor the design of a routing protocol. General criteria are that thenumber of hops on the path or the overall link cost is used as themetric to determine the optimal route.Scalability: The routing protocol should scale well in large wirelessad hoc and sensor networks, with rapid topology changes andlink failures.Control traffic overhead: The routing protocols must minimize the controltraffic overhead required to discover the routes.Efficiency: The routes selected by the routing protocol affect the performanceof the network in terms of delay, throughput, and energyefficiency. Therefore, the routing protocol should aim atimproving the overall network efficiency.In this chapter, the available routing protocols (Clausen and Jacquet 2003,Perkins et al. 2003, Johnson et al. 2003, Park and Corson 1997, Sivakumaret al. 1999, Perkins and Bhagwat 1994, Aceves and Spohn 1999) are reviewed,and an optimal energy-delay routing scheme is introduced. Reactive protocolslike AODV (Perkins et al. 2003), DSR (Johnson et al. 2003), TORA (Parkand Corson 1997), and CEDAR (Sivakumar et al. 1999) compute the routeson-demand, which reduces the control overhead at the cost of increasedlatency. The rapid changes to the topology and link failures present in adhoc networks, and the need to react quickly to the routing demands, makeproactive protocols more suitable for such networks. A few examples ofproactive routing protocols are DSDV (Perkins and Bhagwat 1994), STAR(Aceves and Spohn 1999), and OLSR (Jacquet et al. 2001, Clausen and Jacquet2003). However, these routing protocols attempt to find the minimal hoppath from the source to the destination, an approach that may not providethe optimal path in terms of delay and energy efficiency.